[GHC] #5996: fix for CSE
GHC
ghc-devs at haskell.org
Tue Nov 12 15:23:43 UTC 2013
#5996: fix for CSE
--------------------------------------------+------------------------------
Reporter: michalt | Owner: simonpj
Type: bug | Status: patch
Priority: normal | Milestone: 7.8.1
Component: Compiler | Version: 7.5
Resolution: | Keywords:
Operating System: Unknown/Multiple | Architecture:
Type of failure: Runtime performance bug | Unknown/Multiple
Test Case: | Difficulty: Unknown
Blocking: | Blocked By:
| Related Tickets:
--------------------------------------------+------------------------------
Comment (by simonpj):
Commit message properly formatted
{{{
Fix egregious omission in CSE (Trac #5996)
This patch fixes a bad omission in CSE, thanks to 'michaelt' for spotting
it, and correctly identifying the fix (in cseRhs). The trouble was with
x1 = C a b
x2 = C x1 b
y1 = C a b
y2 = C y1 b
we were not commoning up y2=x2, because we failed to substitute y1:=x1,
so y2's RHS looked different to x2's
I also refactoring, so taht the cs_map in a CSEnv map is
cs_map :: CoreMap (OutExpr, Id)
instead of
cs_map :: CoreMap (OutExpr, OutExpr)
Much nicer!
This doesn't make much difference to allocation, but it gives a
surprisingly
big benefit to binary size.
--------------------------------------------------------------------------------
Program Size Allocs Runtime Elapsed TotalMem
--------------------------------------------------------------------------------
ansi -1.7% -0.8% 0.00 0.00 +0.0%
bspt -1.6% -1.5% 0.01 0.01 +0.0%
cacheprof -1.8% -0.2% +1.6% +1.9% +2.7%
fft -1.4% -1.3% 0.06 0.06 +11.1%
ida -1.4% -1.0% 0.12 0.12 +0.0%
rfib -1.4% -0.1% 0.03 0.03 +0.0%
scs -1.6% -0.1% +1.5% +1.5% +0.0%
spectral-norm -1.3% -0.1% -0.2% -0.2% +0.0%
tak -1.4% -0.1% 0.02 0.02 +0.0%
veritas -1.4% -0.1% 0.00 0.00 +0.0%
--------------------------------------------------------------------------------
Min -2.5% -1.5% -11.8% -11.8% -8.0%
Max -1.0% +0.0% +2.7% +2.5% +11.1%
Geometric Mean -1.3% -0.1% -2.6% -2.6% +0.0%
}}}
--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/5996#comment:10>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler
More information about the ghc-tickets
mailing list